While the predisposition to suicidal behavior is complex, the interpersonal context plays a critical role both as a trigger and a deterrent. Moreover, people who attempt suicide display chronic interpersonal dysfunction and an impulsive and avoidant approach to social problems, prompting questions about the way they make social decisions. Although psychological accounts of interpersonal deficits in suicide exist, these are not integrated with neural mechanisms. Addressing this lacuna, the applicant, a cognitive psychologist, seeks to develop an independent research program in computational psychiatry with a long-term goal of investigating neural mechanisms of social decision-making that may underlie interpersonal dysfunction in suicide and other mental disorders. In the laboratory, suicide attempters display a tendency toward short-sighted, negligent decisions and heightened susceptibility to decision biases. This behavioral profile of decision incompetence is paralleled by disrupted decision-related signals in the ventral prefronto-striatal circuit. The literature and our preliminary studies also suggest that decision-related signals in this circuit are modulated by the social context, which can at times counteract goal-directed, adaptive processing. Thus, the applicant?s short-term aim is to test a conceptual model, wherein decision deficits in suicide attempters result from an interference of automatic responses to the social context with goal-directed processes that subserve adaptive decision-making. The applicant has developed a social exchange paradigm, which examines decision-making under disruptive social influences, modeling one key aspect of the suicidal crisis. Her approach leverages formal learning theory and builds on recent computational studies, dissecting social decision-making as an interaction of automatic and goal-directed processes. The applicant will test the interference hypothesis in two cross-sectional case-control studies of behavior (n=120) and neural decision-related signals (n=60). She will use reinforcement learning (RL) models to contrast behavioral tendencies and neural decision-related signals in depressed suicide attempters with non-suicidal depressed and healthy controls. This approach will elucidate individual differences in the degree to which automatic responses to the social context interfere with goal-directed processes. This research will serve as a platform for interdisciplinary training provided by experts in functional and structural imaging (Aizenstein), suicide phenomenology (Dombrovski and Szanto), social and decision neuroscience (Delgado), cortico-striatal circuitry (Frank), learning theory and computational modeling (Dombrovski and Frank), and personality (Hallquist). The applicant will take advantage of neuroimaging facilities and extensive research infrastructure at the University of Pittsburgh. The proposed study addresses the #1 question of the Prioritized Research Agenda for Suicide Prevention, Why do people become suicidal? Examining learning signals (prediction errors) in a social context and social approach/avoidance, this work applies the RDoC framework (Positive and Negative Valence Systems, Systems for Social Processes) to the study of suicide.